OpenCompass Math - Competition (CompassBench 2604): leaderboard
OpenCompass CompassBench 2604 edition (file dated 2026-05-13): competition mathematics sub-task of the math category. CompassBench is OpenCompass's periodic evaluation of chat models on a private question set; each edition is a separate board. Score 0-100, higher is better.
Metric: Score (%). Source: rank.opencompass.org.cn. Saturation forecast: Around February 2027. 23 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 72.1 | #64 |
| 2 | Kimi K2.6 | 72.1 | #113 |
| 3 | DeepSeek V4 Pro | 71.2 | #112 |
| 4 | Doubao-Seed-2-0-Pro-260215 (High) | 70.8 | |
| 5 | Claude Opus 4.7 (High) | 70.5 | #51 (Claude Opus 4.7) |
| 6 | GPT-5.4 (2026-03-05) (High) | 69.9 | #63 (GPT-5.4 (2026-03-05)) |
| 7 | Claude Sonnet 4.6 (High) | 67.6 | #94 (Claude Sonnet 4.6) |
| 8 | Hy3-preview (High) | 67.6 | #166 (Hy3-preview) |
| 9 | DeepSeek V4 Flash | 67 | #155 |
| 10 | Qwen 3.5 397B A17B | 65.7 | #154 |
| 11 | Qwen 3.6 Max Preview | 65.1 | #91 |
| 12 | Step 3.5 Flash | 63.1 | #281 |
| 13 | GLM-5.1 | 59 | #109 |
| 14 | Gemma 4 31B (IT) | 58 | #178 |
| 15 | Qwen 3.6 27B | 57.7 | #182 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=opencompass-math-competition-compassbench-2604 · How It Works · Data refreshed daily, snapshot 2026-11-05.